Fig. 4

Evaluate SMARTcleaner with H3K4me3 ChIP-seq data. a Numbers of pre- and post-cleaning H3K4me3 peaks. b Change of peak numbers after cleaning. c Numbers of peaks shared or unique to pre-cleaning (“uniqPre”) or post-cleaning (“uniqPost”) data d Overlap of peaks with poly(T/A) sites. e Overlap of peaks with gold standard peaks. f Peaks at the promoter regions (2 kb around TSS). g-j Percentages of peaks with each of the four enriched TF motifs. k Read densities and average profiles for peaks shared by or unique to pre- and post-cleaning data. Reads counts were extracted using seqMINER [33] from 6 million reads randomly sampled from individual samples. Heatmaps were drawn using R package pheatmap, with peaks as row and sorted by read densities. In a-j, each point represents a replicate sample. Mann-Whitney test for b, d-j; t-test for c; ** p < 0.01, n.s., not significant; the comparisons in e-j are either for uniqPre vs uniqPost or uniqPre vs shared